Uli Grasemann, a graduate student in the Department of Computer Science, relied on a synthetic neural network that was specially designed by his adviser, Professor Risto Miikkulainen. Dubbed DISCERN, the network will simulate effects that different types of neurological dysfunction have on human language function.
DISCERN has been changed to prevent it from discarding as much extraneous information, and to put it in layman’s terms, to forget less. This resulted in fantastical and delusional stories, where it will merge elements of unrelated stories.
Ralph Hoffman, a professor of psychiatry at the Yale School of Medicine, claimed that the network’s behavior was eerily similar to that of human schizophrenics. Is that conclusive proof of the hyperlearning hypothesis? Perhaps, but it points toward the potential for using computers in neurological research.